Unsupervised anomaly detection for longitudinal comparison in whole-body PET/CT images
4 affiliations
- 1Department of Computational Diagnostic Radiology and Preventive Medicine, The University of Tokyo Hospital, 7-3-1 Hongo, Bunkyo-Ku, Tokyo, Japan
- 2Department of Radiology, The University of Tokyo Hospital, Tokyo, Japan
- 3Division of Radiology and Biomedical Engineering, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan
- 4Center for Frontier Medical Engineering, Chiba University, Chiba, Japan
Abstract
Purpose: This study investigates the utility of unsupervised anomaly detection for longitudinal comparison of whole-body 18F-fluorodeoxyglucose (FDG)-PET/CT, which (1) reduces false-positive findings compared with subtraction-based methods and (2) enables highlighting newly appearing lesions across diverse anatomical regions without requiring lesion-annotated datasets, unlike supervised approaches.
Methods: A 3D U-Net was trained exclusively on normal PET/CT images to predict voxel-wise probability distribution parameters of the current PET image conditioned on prior PET, prior CT, and current CT images. During anomaly detection, voxel-wise Z-scores were computed from the predicted parameters and used as an abnormality score map. The performance of the proposed method was compared with that of the subtraction method using slice-wise receiver operating characteristic (ROC) analysis and free-response ROC (FROC) analysis.
Results: The area under the slice-wise ROC curve (AUC) was 0.881 for the proposed method and 0.757 for the subtraction method. Region-wise subanalysis demonstrated AUCs of 0.925, 0.918, and 0.824 in the neck, chest, and abdominal regions, respectively, all of which exceeded the corresponding AUCs obtained with the subtraction method. In the FROC analysis, sensitivity at 5.0 false positives per case was 0.728 for the proposed method, compared with 0.375 for the subtraction method.
Conclusions: The proposed method effectively suppressed false-positive findings compared with subtraction imaging and successfully highlighted lesions across diverse anatomical regions without requiring lesion-annotated datasets, as opposed to supervised learning approaches.
Related
- Interleukin pathway genotype and the grandiosity dimension of psychosis: a machine learning gene set analysis with unsupervised anomaly detection controls
- Spatially Context-Aware Transformers Facilitate Modeling-Based Anomaly Detection of Subtle Lesions in Brain MRI Images
- Longitudinal immune trajectories following SARS-CoV-2 vaccination in long-term care facility residents: an unsupervised analysis approach
- Automated corneal endothelial cell detection and segmentation in specular microscopy images using Attention and Faster U-Net
- A unified deep learning framework for multi-lesion detection in diabetic retinopathy using fundus images
BibTeX
@article{nakao2026unsupervised,
title = {{Unsupervised anomaly detection for longitudinal comparison in whole-body PET/CT images}},
author = {Takahiro Nakao and Shouhei Hanaoka and Yukihiro Nomura and Takeharu Yoshikawa and Osamu Abe},
journal = {International Journal of Computer Assisted Radiology and Surgery},
year = {2026},
month = may,
doi = {10.1007/s11548-026-03672-6},
url = {https://doi.org/10.1007/s11548-026-03672-6},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Unsupervised anomaly detection for longitudinal comparison in whole-body PET/CT images
AU - Takahiro Nakao
AU - Shouhei Hanaoka
AU - Yukihiro Nomura
AU - Takeharu Yoshikawa
AU - Osamu Abe
PY - 2026
DA - 2026/05/25
JO - International Journal of Computer Assisted Radiology and Surgery
DO - 10.1007/s11548-026-03672-6
UR - https://doi.org/10.1007/s11548-026-03672-6
AB - Purpose: This study investigates the utility of unsupervised anomaly detection for longitudinal comparison of whole-body 18F-fluorodeoxyglucose (FDG)-PET/CT, which (1) reduces false-positive findings compared with subtraction-based methods and (2) enables highlighting newly appearing lesions across diverse anatomical regions without requiring lesion-annotated datasets, unlike supervised approaches. Methods: A 3D U-Net was trained exclusively on normal PET/CT images to predict voxel-wise probability distribution parameters of the current PET image conditioned on prior PET, prior CT, and current CT images. During anomaly detection, voxel-wise Z-scores were computed from the predicted parameters and used as an abnormality score map. The performance of the proposed method was compared with that of the subtraction method using slice-wise receiver operating characteristic (ROC) analysis and free-response ROC (FROC) analysis. Results: The area under the slice-wise ROC curve (AUC) was 0.881 for the proposed method and 0.757 for the subtraction method. Region-wise subanalysis demonstrated AUCs of 0.925, 0.918, and 0.824 in the neck, chest, and abdominal regions, respectively, all of which exceeded the corresponding AUCs obtained with the subtraction method. In the FROC analysis, sensitivity at 5.0 false positives per case was 0.728 for the proposed method, compared with 0.375 for the subtraction method. Conclusions: The proposed method effectively suppressed false-positive findings compared with subtraction imaging and successfully highlighted lesions across diverse anatomical regions without requiring lesion-annotated datasets, as opposed to supervised learning approaches.
SN - 1861-6410
SN - 1861-6429
ER -